The present research attempts to enhance our understanding of the formation, retention, and utilization of complex, ill-defined categories. One approach is to isolate variables influencing the formation of categories, and to understand how abstracted information is utilized in perception and retrieval. A second approach focuses upon the inter-relationships among multiple categories and their members as a function of learning variables. Specifically, can one 'track' the genesis and modification of a conceptual space via the method of multidimensional scaling by judicious manipulation of learning variables? The hypothesized conceptual space is viewed as proceeding from a relatively undifferentiated state to one in which the final configuration mirrors the requisite category learning. Variables such as within-category experience, number of contrasting categories, distortion of stimulus members, and varying time delays are suggested to be important determinants in the development of such a spatial model. With sufficient time delays, it is proposed that the configuration may show diffusion from its learned state. Measurements such as the degree of category structure, the extent to which the abstracted prototype occupies the ideal (geometric) center of its exemplars, and the relative clustering for categories of different size are envisioned as providing insight into the genesis and decay of conceptual information.